Cellular radio services have been extremely successful in providing untethered voice communications. With the advent of new personal communications services, wireless access is expected to become even more popular. At the same time, personal computers and Internet services are experiencing explosive growth due to low cost, high performance computer technologies and attractive network applications. The popularity of the World Wide Web is further accelerating the explosive growth of personal computers and Internet services. Advanced Cellular Internet Services (ACIS) are targeted for applications, such as web browsing, which have a peak downlink data rate on the order of 1-2 Mb/s using a wide-area cellular infrastructure.
A major obstacle for wireless data services is the capability of cellular networks for providing a transporting bandwidth sufficiently high for meeting the needs of web browsing and information exchange applications. For example, the bandwidth requirements for two-way Internet radio links are expected to be highly asymmetric. That is, a peak downlink data rate of about 2 Mb/s is expected, with an uplink peak data rate of about one to two orders of magnitude less. Further, since there is a need for providing both voice and data services using a limited frequency spectrum, the algorithm used for channel assignment must aggressively reuse frequency to within a fraction of the total available bandwidth.
Further, ACIS poses several technical challenges in the area of medium access control (MAC) methodology. A suitable MAC method for ACIS must address not only packet server allocation (base station, antenna sector, radio transceiver and time slot), but also channel assignment for a given server. That is, a suitable MAC protocol must consist of two parts, an access protocol, and a channel assignment algorithm that handles the asymmetric uplink/downlink bandwidth expectations and the aggressive frequency reuse issues, respectively. The access protocol part of the MAC protocol is similar to a conventional wireline-based access protocol in that it must avoid “hard” collisions in the same cell, while the channel assignment part must avoid “soft” collisions among different cells caused by co-channel interference.
Further still, a suitable ACIS MAC protocol must also handle diverse types of multimedia traffic having different Quality of Service (QoS) requirements. For example, delay-sensitive traffic, such as packet voice traffic, and error-sensitive traffic, such as packet data traffic ranging from signaling a simple command to downloading a large image file, must be assigned radio resources that meet specific requirements. Accordingly, when packet access techniques are used, radio resources, that is, channels, are assigned only when there are packets to be delivered. The price of the advantages obtained by this statistical multiplexing is a rapidly changing interference environment. Thus, a circuit-access channel assignment method that is based on continuous interference measurement and averaging may not achieve desirable results in a packet switched system because the interference environment measured may be significantly changed when actual transmissions occur, such as that disclosed by M. Frullone et al., “PRMA Performance In Cellular Environments with Self-Adaptive Channel Allocation Strategies,” IEEE Trans. Veh. Tech., November 1996, pp. 657-665. The time delay occurring between measurement and transmission, known as “setup latency”, has detrimental effects on system performance unless some “inertia” is introduced for reducing interference variations.
Existing wireless data rates of up to several tens of kb/s may be more than an order of magnitude short of what is required to make popular web browsing and information exchange applications user-friendly when used with wireless access. For example, see J. F. DeRose, “The Wireless Data Handbook,” Quantum Publishing, Inc., 1994. Although there are approaches that propose providing up to about 100 kb/s peak rates using second-generation digital cellular technologies, such as the General Packet Radio Service (GPRS) for the GSM system, such approaches usually aggregate several traffic servers (e.g., multiple time slots in TDMA systems) as a single high-rate data server using conventional frequency reuse strategies. See, for example, R. Ludwig, “Downlink Performance of the General Packet Radio Service for GSM,” Proceedings, 3rd International Workshop on Mobile Multimedia Communications (MoMuC3), Paper # A.2.2.3, September 1996. Approaches that use multiple time slots for providing a single high-rate data server consume significant portions of system capacity and would seriously compromise traditional voice services when high-speed wireless data services become popular.
Presently, there are wireless Ethernet modems that provide several Mb/s or higher peak data rates for local environments. For example, see A. R. Nix, “Hiperlan Compatible Modulation and Equalisation Techniques—What are the real choices,” ETSI RES-10 standard contribution RES-10TTG/93/78, December 1993. The feasibility of even higher data rates has been demonstrated by some wireless ATM hardware prototypes. Nevertheless, peak data rates of several Mb/s have not been demonstrated in wide-area cellular networks due to significant path loss and delay spread encountered in a cellular network environment. Furthermore, a high bandwidth requirement makes frequency reuse much more challenging in a cellular environment because the limited available spectrum cannot be easily divided into a large number of reuse channel sets.
Recently, simulations of a modulation method employing Orthogonal Frequency Division Multiplexing (OFDM), antenna diversity and channel coding have shown promise for providing 1-2 Mb/s peak downlink rates in an ACIS environment. For example, see L. J. Cimini and N. R. Sollenberger, “OFDM with Diversity and Coding for High-Bit-Rate Mobile Data Applications,” Proceedings, 3rd International Workshop on Mobile Multimedia Communications, Princeton, September 1996, paper # A.3.1.1. While a number of challenges remain and a hardware implementation of a low cost, robust modem is difficult, the transmission technique appears to be feasible.
Among the useful attributes of existing approaches, interference sensing used by a measurement-based DCA approach and knowledge of preferred channels used by a channel segregation (CS) technique are advantageous for packet channel assignment. Interference sensing enables base and/or mobile stations to determine potential interference before choosing a given channel and avoiding inter-cell collisions of packets that are delivered using the same radio channel. For an example an interference measurement-based DCA approach, see J. Chuang, “Performance Issues and Algorithms for Dynamic Channel Assignment,” IEEE J. Select. Areas Comm., August 1993, pp. 955-963. Channel segregation divides a radio spectrum into different groups of preferred channels through a learning process, thus preventing adjacent radio transceivers from simultaneously using the same channels. Channel assignments are made robust even when interference sensing is too slow for detecting rapid variations in packet time scale. For details regarding the original channel segregation technique, see F. Furuya et al., “Channel Segregation, A Distributed Adaptive Channel Allocation Scheme for Mobile Communications Systems,” Proceedings, 2nd Nordic Seminar on Digital Land Mobile Radio Communications, October 1986, also appearing in IEICE Trans., Vol. E74, June 1991, pp. 1531-1537.
The original CS algorithm increments a priority value of a channel if the channel is successfully assigned for use and does not experience interference after communications begin. After a period of initial trial-and-error, an entire system using the original CS algorithm autonomously forms a reasonable reuse plan. The original CS algorithm is a simple algorithm, but would be more effective if failures could be reduced during the learning phase.
Without introducing memory, measurement-based DCA using two-way interference sensing is effective for circuit access even with significant setup latency. For example, see J. C.-I. Chuang and N. Sollenberger, “Performance of Autonomous Dynamic Channel Assignment and Power Control for TDMA/FDMA Wireless Access,” IEEE Journal on Selected Areas in Communications, special issue on Wireless and Mobile High Speed Communication Networks, October 1994, pp. 1314-1323. A small performance degradation experienced by such an approach is the result of a non-zero probability of “soft collisions” that occur because more than one packet from nearby base stations are delivered on the same channel. As latency rises, the collision probability increases as more newly selected links are “blind” until communications begin. For packet access, duty cycle of channel usage is shortened due to resource sharing among multiple mobiles. This results in higher interference fluctuation and more frequent channel assignment. Therefore, good averaging during measurement is more difficult to achieve and the latency effect is expected to be more pronounced. However, strong non-blind interferers should still be detectable (and thus be avoidable) with a fast measurement. For measurement-based DCA to be effective in packet access, it is thus important to either reduce measurement time and avoid undue setup latency and/or make the assignment robust to setup latency. To this end, selecting channels according to a priority order reduces the impact of soft collisions even though measurements are blind during setup latency. This is because priority orders for the neighboring base stations are likely to be different if a proper ranking/updating algorithm is employed for “segregating” channels during a learning process.
Examples of existing wide-area wireless packet data services include Advanced Radio Data Information Service (ARDIS), RAM Mobile Data, Cellular Digital Packet Data (CDPD) and second-generation digital cellular networks. These conventional services generally offer raw data rates lower than 28.8 kb/s and do not reuse spectrum efficiently for packet data. Typically, wireline-based data MAC protocols are enhanced by separate radio resource management functions that perform channel assignment.
CDPD provides packet access by using cellular infrastructure. For example, see Cellular Digital Packet Data, System Specification & Implementor Guidelines, CD-ROM, Release 1.1, Jan. 19, 1995, CDPD Forum. A channel sniffing and hopping process is used for selecting idle voice channels for access. A Digital Sense Multiple Access/Collision Detection (DSMA/CD) protocol is then used for sending digital busy signals for intra-cell contention management. The basic concept is similar to the Carrier Sense Multiple Access/Collision Detection (CSMA/CD) protocol used in wireline-based LANs and the Carrier Sense Multiple Access/Collision Avoidance (CSMA/CA) protocol used in wireless LANs. No additional frequency reuse method is employed for managing inter-cell contention except for conventional frequency planning, such as is typically performed for voice services using fixed channel assignment (FCA) methods. Thus, the CDPD approach is not sufficient for ACIS applications because a large bandwidth for individual packet channels is used.
Similar to CDPD, both RAM and ARDIS use protocols that are similar to CSMA for packet access contention. Unlike CDPD, though, RAM and ARDIS both have dedicated spectra and dedicated networks for packet data communications. For example, see RAM Mobile Data System Overview, July 1996, RAM Mobile Data. The frequency reuse aspect of the RAM MAC protocol is conventional FCA and, consequently, not efficient for the ACIS environment. The ARDIS approach is similar to RAM Mobile DATA in the services offered, but the ARDIS frequency reuse scheme focuses more on providing coverage by trading off capacity. At times, more than one base station is allowed to use the same channel for reducing coverage holes. Thus, the ARDIS approach is not efficient enough for the ACIS environment.
Recently, there are research and standardization efforts for the second- and third-generation digital cellular systems for providing packet data access. As previously mentioned, the second-generation system approaches usually aggregate several traffic servers for achieving high data rates and are not suitable for the ACIS environment. On the other hand, the third-generation systems and associated research activities, e.g., the European activities on Advanced Communications Technologies and Services (ACTS) are beginning to address high-rate packet access. For example, see IEEE Personal Communications Magazine, “Special Issue on The European Path Towards UMTS,” February 1995. Among initiatives of ACTS, the Advanced TDMA (ATDMA) system is of particular interest to the ACIS environment. A relatively mature proposal of the ATDMA MAC protocol is the PRMA++ protocol in which specific slots for reservation, fast paging and acknowledgement are included in every frame for performing reservation and release of packet traffic servers. Details are disclosed by J. Dunlop et al., “Performance of a Statistically Multiplexed Access Mechanism for a TDMA Radio Interface,” IEEE Personal Communications Magazine, June 1995, pp. 56-64. The PRMA++ protocol enhances the original Packet Reservation Multiple Access (PRMA) disclosed by D. Goodman et al., “Packet Reservation Multiple Access for Local Wireless Communications,” IEEE Trans. on Commun., August 1989, pp. 885-890, by reducing or eliminating intra-cell traffic packet collisions and accelerating the setup process.
F. Borgonovo et al. disclose a similar approach by using slots with different functionalities. A hybrid polling and reservation scheme is introduced that further reduces contention of uplink traffic by allowing a base station to poll its active mobiles. See F. Borgonovo et al., “Capture-Division Packet Access for Wireless Personal Communications,” IEEE Journal on Selected Areas in Communications, May 1996, pp. 609-622, and F. Borgonovo et al., “Capture-Division Packet Access: A New Cellular Access Architecture for Future PCNs,” IEEE Communications Magazines, September 1996, pp. 154-162. Frequency reuse is achieved by capture effects and a flexible frame structure is adaptively formed based on the varying rates of polled users with different QoS requirements. The approach does not use a conventional TDMA frame structure and, consequently, was termed “Capture Division Packet Access (CDPA).” However, both PRMA++ and CDPA do not specify a detailed channel assignment algorithm and methods for effective frequency reuse in packet access remains an area of active research worldwide. For example, see M. Frullone et al., supra; A. Baiocchi et al., “The Geometric Dynamic Channel Allocation as a Practical Strategy in Mobile Networks with Bursty User Mobility,” IEEE Transactions on Vehicular Technology, February 1995, pp. 14-23; A. Srivastava and J. Chuang, “Access Algorithms for Packetized Transmission in the presence of Co-channel Interference,” IEEE VTC '96, Atlanta, Ga., April 1996, pp. 121-125; A Srivastava and J. Chuang, “Packetized Wireless Access for Data Traffic in Frequency Reuse Environments,” Seventh International Symposium on Personal Indoor Mobile Radio Communications (PIMRC), Taipei, Taiwan, October 1996, pp. 1150-1154; and T. Benker, “Dynamic Slot Allocation for TDMA-Systems with Packet Access,” Multiaccess, Mobility and Teletraffic for Personal Communications,” B. Jabbari, P. Godlewski and X. Lagrange (Editors), Kluwer Academic Publishers, 1996, pp. 103-116.
The Frullone et al. publication, supra, and both Srivastava et al. publications, supra, consider the effects of co-channel interference using the PRMA protocol and propose methods for trading signal quality for traffic throughput. Both groups of researchers considered choosing only a subset of available channels at a base station for contention, with a larger subset of available channels resulting in a better throughput with possible quality degradation than with a smaller subset of available channels. Frullone et al. employed channel segregation, initially proposed by Furuya et al., supra, for prioritizing all available channels through a learning process that is based on the probability of access success, that is, no hard collisions caused by intra-cell contention and no soft collisions caused by co-channel interference during a talkspurt. After a certain period of training time, adjacent base stations tend to have different sets of preferred channels, thus achieving adaptive frequency reuse. The Frullone et al. approach improves frequency reuse even under the rapidly changing interference of the packet access environments. Nevertheless, the disadvantage of this approach is that it is not easy to avoid interference during the learning process, i.e., bad channels are avoided only after failure occurs.
Srivastava et al., supra, propose that a base station broadcasts a subset of channels having the lowest interference for all associated mobile stations for contending access based on continuously sensing uplink interference. This is realized by associating with each channel a “permission probability.” Highly interfered-with channels are given zero permission probability, thus preventing such channels from been used. Fuzzy logic is used for determining the permission probability based on the prevailing interference conditions and the number of idle slots at the base station. An access request is sent by a mobile station only when a locally generated random number is lower than the permission probability. This approach achieves good delay and throughput characteristics for both packet voice and data applications using a simple set of intuitive control rules. However, only the uplink access issue is addressed, which is less challenging in ACIS applications because of the asymmetrical uplink/downlink bandwidth requirements. Additionally, downlink transmission is assumed to be continuous for facilitating rapid access, possibly requiring more spectrum than that available in an ACIS environment because continuous downlinks represent 100% downlink spectral usage and results in high interference levels.
To improve signal quality on both links, J. Chuang, “Performance Issues and Algorithms for Dynamic Channel Assignment,” IEEE J. Select. Areas Comm., August 1993, pp. 955-963, discloses a two-way dynamic channel assignment algorithm for circuit-access environments. A pilot-based scheme and its frame structure were proposed for implementing this approach with low latency and no blind slots. Also see J.C.-I. Chuang et al., “A Pilot Based Dynamic Channel Assignment Scheme for Wireless Access TDMA/FDMA Systems,” The International Journal of Wireless Information Networks, Vol. 1, No. 1, January 1944, pp. 37-48. Similar to the approach disclosed by Srivastiva et al., “Access Algorithms for Packetized Transmission in the presence of Co-channel Interference,” IEEE VTC '96, Atlanta, Ga., April 1996, pp. 121-125, the J.C.-I. Chuang et al. approach requires base stations to continuously monitor the uplink and broadcast lists of good idle channels on a control frequency, thereby avoiding continuous downlink transmission on the traffic channels. Upon receiving the good idle channel list, a mobile station measures interference on the candidate channels and selects a channel with sufficiently low interference, thus achieving good quality in both directions. This requires significant measurement and feedback transmission by the mobiles, but the advantages are significant. For the ACIS environment, however, it is not clear whether the required functionalities can be performed fast enough for taking full advantage of this approach. Furthermore, a control scheme must be devised for providing a mechanism for logical pairing or feedback for both directions, and which must be performed for typically asymmetric two-way traffic.
An adaptive array beamformer is a device that separates signals collocated in the frequency band but separated in the spatial domain. This provides a means for separating a desired signal from interfering signals. The adaptive array beamformer automatically optimizes the array pattern by adjusting the elemental control weights until a prescribed objective function is satisfied. A specially designed algorithm provides the means for achieving the optimization. An adaptive array is beneficial for a modern wireless communication system because it has the advantage of increasing the power available to the desired receiver. This is particularly beneficial for a code division multiple access (CDMA) system because the interference received by another receiver in an adjacent area is significantly reduced relative to conventional omni-directional or sectored transmissions.
An adaptive array includes a collection of transmitters and receivers coupled to a corresponding element of the antenna array. The received beamform is determined by measuring the power and relative phase of a desired signal received on each antenna element and its corresponding receiver. The transmit beam is formed by varying the relative phase and power of each signal transmitted by the multitude of transmitters and corresponding antenna elements.
An adaptive array includes a collection of antenna elements. A weighting algorithm processes the signal received by each antenna element to produce a weighted signal. The weighted signals from the antenna elements are synthesized to form a synthesized pattern. The weighting algorithm is chosen to achieve a specific purpose or amount of weighting. For example:                1. Least-Mean-Squared (LMS)—A method in which the weighting is determined so as to minimize the root error component of a received signal with reference to a known reference signal;        2. Maximum Signal-to-Noise Ratio (MSN)—A method in which the Signal-to-Noise Ratio (SNR) is maximized with a known direction of a desired wave;        
3. Power Inversion (PI)—A method in which a strong jamming wave signal is suppressed by the minimization of received power in, for example, reception of a frequency diffusion modulation signal; and                4. Constant Modulus Algorithm (CMA)—A method in which an interference wave is suppressed by utilizing the constant envelope property of an amplitude component of a frequency-modulated or a phase-modulated signal, are proposed.        
The adaptive array beamforming process is simplified in a time division duplex (TDD) system because the communications to and from an adaptive array occur on a common frequency but during a different time slot. This enables the adaptive array to create a transmit beam pattern substantially equal to a beam pattern received by the adaptive array. In a frequency division duplex (FDD) system, however, transmissions and receptions occur on different frequencies, having different propagation paths. The beamform of a TDD system is more accurately formed because the transmit and receive propagation characteristics are substantially the same.
What is needed is a wireless communication system that includes an adaptive array in the receiver, the transmitter, or both. By taking advantage of the benefits of an adaptive array, a wireless communication system will improve the performance in the channel selected. With the prior art, channel selection is based on the interference level of each channel after adaptive array combining, and since adaptive arrays suppress interference, the wireless communication system with adaptive arrays using the prior art optimizes the performance of a given link by selecting the channel with the lowest interference. However, an adaptive array can substantially suppress interferers, but only when the number of interferers is less than the number of antennas. Thus, with the adaptive array the channel selection process of the prior art can place interferers close together. If this results in too many interferers on another link, the performance of that link may be seriously degraded. As a result, the above channel selection process can result in overall system performance that is worse than without an adaptive array.